Character Legibility versus Resolution in Image Processing of Printed Matter

Abstract

In order to determine optimum resolution requirements for digital facsimile systems, a legibility study was conducted. The study was designed to measure the quality of output copy as a function of equipment parameters and copy parameters. The experimental equipment permitted systematic variations in horizontal and vertical resolution, scan direction, and simulated transmission noise; the test documents prepared included uppercase and lowercase characters of several sizes. During the course of the overall study, more than 300 subjects read the facsimile output copies; approximately 750 000 responses were accumulated and analyzed. The data are summarized by showing the spatial resolutions required to maintain 97.5 percent legibility for varying character sizes. Measurements were also made on secondary parameters, indicating slight differences in legibility due to scan direction, and a decrease in legibility when type was lowercase rather than uppercase. The decrease in legibility was also measured for a model of channel errors that assumes the use of run-length compression coding. Some of the results of this study can be generalized to other display systems.